You are on page 1of 37

Ethnic Population Transfers and

Partitions in the Twentieth Century

U n t i l recently, there
has been a near consensus among policymakers and scholars that the objective
of ethnic conflict management should be to support and preserve integrated,
multiethnic societies. In the last few years, however, the idea that separating
the warring populations may be the best solution to many of the most intense
ethnic conflicts has been gaining ground. Events in Bosnia have supported this
trend, as observers note that the more the warring groups have separated, the
more peaceful their relations have become, while proposals to thoroughly
reintegrate them command less and less support.' In addition, a growing body
of scholarship that focuses on the role of intergroup security dilemmas in
ethnic conflicts argues that intermixed population settlement patterns can
promote escalation of violence, implying that separation of warring groups
may dampen conflict2
~

~~

~~

Chaim D. Kaufmann is Associate Professor of International Relations at Lehixh University.


The author's thanks are owed to Robert Art, Pauline Baker, John Mearsheimer, Robert Pape,
Edward Rhodes, Jack Snyder, Monica Toft, Stephen Van Evera, Barbara Walter, and the members
of the University of Chicago Program on International Security and Policy for comments. Research
for this article was supported by the United States Institute of Peace.
1. John J. Mearsheimer, "Shrink Bosnia to Save It," Nmu York Times, March 31, 1993; Mearsheimer
and Stephen W. Van Evera, "When Peace Means War," New Republic, December 18,1995, pp. 16-21;
Robert M. Hayden, "Schindler's Fate: Genocide, Ethnic Cleansing, and Population Transfers," Slaziic
Review, Vol. 55, No. 4 (Winter 1996), pp. 740-742; Ivo H. Daalder, "Bosnia after SFOR: Options for
Continued U.S. Engagement," Survival, Vol. 39, No. 4 (Winter 1997-98), pp. 5-18; Robert A. Pape,
"Partition: An Exit Strategy for Bosnia," Survival, Vol. 39, No. 4 (Winter 1997-98), pp. 25-28; and
Michael OHanlon, "Turning the Cease-fire into Peace," Bruokings Review, Vol. 16, No. 1 (Winter
1998), pp. 4144. In addition, some analysts who oppose the partition of Bosnia admit that
reintegration of the separated populations would be very difficult. See Charles G. Boyd, "Making
Bosnia Work," Foreign Affairs, Vol. 77, No. 1 (January/February 1998), pp. 42-55; Susan L. Woodward, "Avoiding Another Cyprus or Israel," Brookings Review, Vol. 16, No. 1 (Winter 1998),
pp. 4548; and Jane M.O. Sharp, "Dayton Report Card," International Security, Vol. 22, No. 3 (Winter
1997/98), p. 133. Flora Lewis, "Reassembling Yugoslavia," Foreign Policy, No. 98 (Spring 1995),
pp. 132-144, argues that Bosnia could be reintegrated.
2. Barry R. Posen, "The Security Dilemma and Ethnic Conflict," in Michael E. Brown, ed., Ethnic
ConfZict and Internalfond Security (Princeton, N.J.: Princeton University Press, 1993), pp. 103-124;
Chaim Kaufmann, "Possible and Impossible Solutions to Ethnic Civil Wars," International Security,
Vol. 20, No. 4 (Spring 1996), pp. 136-175; and Daniel L. Byman, "Divided They Stand: Lessons
about Partition from Iraq and Lebanon," Security Studies, Vol. 7, No. 1 (Autumn 1997), pp. 1-29.
See also Myron S. Weiner, "Bad Neighbors, Bad Neighborhoods: An Inquiry into the Causes of
Refugee Flows," Iwtemntionul Security, Vol. 21, No. 1 (Summer 1996), pp. 37-38; and Clive J.
Christie, "Partition, Separatism, and National Identity," Polifical Quarterly, Vol. 63, No. 1 (JanuaryIntrmtzoiial Security, Vol. 23, No. 2 (Fall 1998), pp. 120-156
0 1998 by the President and Fellows of Harvard College and the Massachusetts Institute of Technology

120

When All Else Fails 121

Separating populations, however, remains deeply controversial. Even when


carried out safely, population transfers inflict enormous suffering, including
loss of homes and livelihoods and disruption of social, religious, and cultural
ties. Thus they can be justified only if they save the lives of people who would
otherwise be killed in ethnic violence. Critics argue that ethnic population
transfers, and the partitions that often accompany them, generally do not
reduce suffering and death but actually increase them.
The most important empirical evidence marshaled against demographic
separation rests on the outcomes of four famous twentieth-century partitionsIreland, India, Palestine, and Cyprus-all of which were accompanied by
large-scale population transfers and by substantial ~ i o l e n c e . ~
The question addressed in this article is: If the logic of demographic separation is correct, why were the partitions and population transfers in these four
cases so violent? There are three possibilities: the violence in these cases could
be evidence that the theory is wrong; the violence could have resulted from
idiosyncratic factors that do not shed light on the causal logic of the theory; or
the pattern of violence in these cases could be just as predicted by the theory,
which would mean that violence in these cases (and future cases) could probably be reduced if policymakers facing severe ethnic conflicts were more
willing to consider the option of separating hostile populations.
To answer this question, I investigate the records of these four cases and find
that the critics claims are not justified. In all four cases separation of the
warring groups did reduce subsequent violence. Continuing or resurgent intergroup violence in limited regions within some of the cases has resulted not
from partition or from separation but rather from the incompleteness of separation of the hostile groups in those specific areas.
This article is divided into three sections. The first assesses the state of the
debate on demographic separation as a remedy for ethnic wars and identifies
the empirical questions that must be answered in order to advance it. The
second section investigates whether the net effects of the partitions and population transfers in the Irish, Indian, Palestinian, and Cypriot cases were to
reduce loss of life or to increase it. The third section considers whether partitions and population transfers create undemocratic states.
March 1992), pp. 6&78. On why separation can resolve ethnic conflicts but not ideological civil
wars, see Chaim Kaufinann, Intervention in Ethnic and Ideological Civil Wars, Security Studies,
Vol. 6, No. 1 (Autumn 19961, pp. 62-103.
3. Radha Kumar, The Troubled History of Partition, Foreign Affairs, Vol. 76, No. 1 Uanuary/February 1997), pp. 22-34.

International Security 23:2 1 122

The State of the Debate


This section lays out the sides in the current debate and the requirements for
advancing our knowledge on the effects of population transfers and partitions.
THE CASE FOR SEPARATION

Whenever ethnic communities cannot rely on a strong and impartial central


state to prevent civil strife, all groups must mobilize for self-defense. However,
the material and rhetorical measures that groups use to mobilize for defense
also pose offensive threats to other groups, creating a security dilemma in
which no group can provide for its own security without threatening the
security of others. The intensity of this security dilemma is in part a function
of demography: the more intermixed the pattern of settlement of the hostile
populations, the greater the opportunities for offense by either side; and it
becomes more difficult to design effective measures for community defense
except by going on the offensive preemptively to cleanse mixed areas of
members of the enemy group and create ethnically reliable, defensible

enclave^.^
The same dynamic also prevents de-escalation of ethnic wars until or unless
the warring groups are substantially separated (or one side conquers or annihilates the other). Solutions that aim both to restore multiethnic civil politics
and to avoid population transfers, such as institution building, power sharing,
and identity reconstruction, cannot work during or after an ethnic civil war
because they do not resolve the security dilemma created by mixed demography5 As long as both sides know that the best security strategy for each is to
engage in offense and in ethnic cleansing, neither can entrust its security to
hopes for the others restraint.6The policy implication is that the international
4. Posen, Security Dilemma and Ethnic Conflict, pp. 108-111.
5. The processes of war, especially reports of real or imagined enemy atrocities, also harden ethnic
identities and solidify hostility and mistrust, creating additional hard-to-counter threat perceptions
even in excess of real threats; this effect persists for a considerable time even after the end of
large-scale fighting. Kaufmann, Possible and Impossible Solutions, pp. 141-145, 150-151.
6. For additional types of proposed solutions to ethnic conflicts, see Donald L. Horowitz, Making
Moderation Pay, in Joseph V. Montville, ed., Conflict and Peacemaking in MuItiethnic Societies (New
York Lexington Books, 1991), pp. 451476; Arend J. Lijphart, The Power-Sharing Approach, in
ibid., pp. 491-510; Gidon Gottlieb, Nation against State (New York: Council on Foreign Relations,
1993); and I. William Zartman, Putting Things Back Together, in Zartman, ed., Co//apsed States:
The Disintegration and Restoration of Legitimate Authority (Boulder, Colo.: Lynne Rienner, 19951,
pp. 267-273. For an analysis that focuses on perceptual rather than structural aspects of intergroup
security dilemmas, and recommends solutions based on institution and confidence building, see
David A. Lake and Donald Rothchild, Containing Fear: The Origins and Management of Ethnic
Conflict, International Security, Vol. 21, No. 1 (Fall 1996), pp. 41-75.

When All Else Fails 1 123

community should endorse separation as a remedy for at least some communal


conflicts; otherwise, the processes of war will separate the populations anyway,
at much higher human cost.
The critical causal factor is separation of people into defensible enclaves, not
partition of ~overeignty.~
Conversely, partition without separation only increases conflict, as it did in Croatia and Bosnia in 1991-92.
THE CASE AGAINST SEPARATION A N D PARTITION

Among most international organizations, Western leaders, and scholars,


population transfers and partition have long been dirty words. With rare
exceptions, the United Nations has supported states against secession movements, and the UN High Commission for Refugees (UNHCR) prefers to bring
safety to people, not people to safety.8Opponents argue that ethnic partitions
and population transfers have three main flaws: (1) rather than dampening
violence, partitions and population transfers actually cause violence; (2) they
generate new conflicts, often by transforming c i v i l conflicts into international
ones; and (3) partitions create rump states that are undemocratic and culturally
narrow, perpetuating intercommunal hatred. The first two criticisms are the
most serious because they concern the central issue of whether demographic
separation saves or costs human lives. The third is important because it suggests that refugees may find themselves in a polity more repressive than the
one they left.1

7. Although, in principle, final political arrangements could be based on either regional autonomy
or separate sovereignty, in practice, demographic separation is likely to be accompanied by
partition, for three reasons. First, one side will often insist on partition. Second, whenever the
international community intervenes to facilitate population transfers, it will need to speclfy partition lines, whether these are between what are to become autonomous provinces or between
independent states. Third, because international law favors sovereign states, granting sovereignty
will usually improve a groups ability to maintain its security.
8. UNHCR, Working Document for the Humanitarian Issues Working Group of the International
Conference on the Former Yugoslavia (Geneva: UNHCR, 1992). According to former UN Secretary-General Boutros Boutros-Ghali, The new danger which will appear in the world in the next
ten years is more kagmentation. . . . We will not be able to achieve any kind of economic
development, not to mention more disputes on boundaries. UN Chief Fears World Could Split
into 400 Mini-states, Montreal Gazette, September 21, 1992. On international practice toward
secession movements, see Lee C. Buchheit, Secession: The Legitimacy of Self-Determination (New
Haven, Conn.: Yale University Press, 1978).
9. Robert Schaeffer, Warpaths: The Politics of Purtition (New York: Hill and Wang, 1990); Amitai
Etzioni, The Evils of Self-Determination, Foreign Policy, No. 89 (Winter 1992-93), pp. 21-35;
Gidon Gottlieb, Nations without States, Foreign Affairs, Vol. 73, No. 3 (May/June 1994), pp. 100112; and Kumar, Troubled History of Partition.
10. Additional criticisms of partition and population transfers include claims that they damage
prospects for future economic growth and that international support for one instance could create

International Security 23:2 124

The implication of the critics' logic is that most of those who have become
refugees in ethnic conflicts could have safely remained in their original homes
and enjoyed reasonable economic, political, and cultural freedom if partition
and population transfers had not been externally imposed on them. Their
bottom line is that working to reintegrate ethnic groups at war with each other
is both more moral and, in the long run, more practical than acquiescing in
partition.
This is wrong. The security dilemma generated by intermixed populations
ensures that ethnic wars always separate the warring communities; this process cannot be stopped except by permanent military occupation or genocide,
or by not having the war in the first place. When ethnic conflicts turn violent
they generate spontaneous refugee movements, for several reasons: people are
afraid to stay in areas where ethnic fighting is ongoing or expected to begin,
or they are forced to leave by their neighbors, marauding gangs, or a conquering army. Thus the question in the midst of severe ethnic conflict is not whether
the groups will be separated but how-with protection, transport, subsistence,
and resettlement organized by outside powers or institutions, or at the mercy
of their ethnic enemies and of bandits? Refusal or failure to organize necessary
transfers does not protect people against becoming refugees, but inflicts disaster on them when they do.
The critics' charge that partitions and population transfers create illiberal
states is also misguided." Although it is true that not all partition successor
states are liberal democracies, the successor states created by the partitions
studied in this article are not less democratic than either their predecessors or
their neighbors. Even though several of the successor states have discriminatory laws, such discrimination is generally less intense than what the prepartition minorities would llkely have faced under majority rule.
REQUIREMENTS FOR RESOLVING THE DEBATE

The maximum universe relevant to this debate would be all cases of border
changes or population movements that altered the ethnic makeup of one or
more states. In practice, this universe would be uncountably large, so for this
a "moral hazard that would encourage proliferation of secession movements. These issues are
considered in Kaufmann, "Possible and Impossible Solutions," pp. 170-173.
11. The claims of both sides in this debate about effects on political development are best understood as "other things being equal," given that political development is also affected by numerous
nonethnic factors, such as economic development, income distribution, preexisting political institutions, and so forth.

When All Else Fails I 125

study I define a more manageable set of the cases most relevant to the policy
utility of demographic separation today: specifically, all twentieth-century ethnic partitions and secessions that have led to the formation of new states,
roughly twenty in a11.12Table 1 summarizes this set.
Within this set, there is a strong association between high violence and large
population transfers; almost all of the high-violence cases involve more substantial refugee movements than any of the low-violence cases. This outcome,
however, is consistent with the arguments of both sides in the debate: proponents of demographic separation contend that high violence causes population
movements, while critics contend the reverse.
To resolve this debate over cause and effect, the cases that are most important to investigate in detail are the high-violence partitions. Because both sides
in the debate are most concerned with the causes of extreme violence, the
low-violence cases cannot be decisive. We should also focus on partition rather
than secession cases, given that our purpose is to assess whether international
intervention reduces or increases the costs of ethnic conflict. This article studies
four of the five cases that qualify-Ireland, India, Palestine, and Cyprus-and
that also have the additional virtue of being the same cases most commonly
used by the critics.

Separation and Violence


This section examines the records of the partitions of Ireland, India, Palestine,
and Cyprus to judge the validity of the claims (1) that separation and partition
increase rather than reduce short-run violence, and (2) that they also perpetuate or actually increase intergroup hatred and violence in the long run.13

12. I define partitions as separationsjointly decided upon by the responsible powers: either agreed
between the two sides (and not under pressure of imminent military victory by one side), or
imposed on both sides by a stronger third party. Secessions are new states created by the unilateral
action of a rebellious ethnic group.
We could study ethnic population transfers between states to see whether they reduce subsequent interstate violence. Indeed, the records of the two largest ethnic population transfers in
twentieth-century Europe-the Greco-Bulgarian-Turkish population exchanges in the 1920s and
the expulsion of ethnic Germans from Eastern Europe after World War 11-suggest that they may.
Each of these exchanges was preceded by a series of wars that cost many times more lives than
the population transfers did, and each has been followed by interstate peace.
13. The following discussions pay special attention to the roles of the minority communities in
each case, because it is usually minorities who are most concerned about their group's security
and who press hardest for partition.

International Security 23:2 126

~~

Table 1. Ethnic Secessions and Partitions, 1900-94.a


Secessions

Partitions

Low
Violence

Baltic states, 1918


Finland, 1918
Soviet Union, 1990-91
Slovenia, 1991
Macedonia, 1992

Norway, 1905
Austria-Hungary, 1918-19b
Ottoman Empire, 1918-19c
Singapore, 1965
Slovakia, 1993

High
Violence

Algeria, 1962
Bangladesh, 1971
Nagorno-Karabakh, 1991*
Iraqi Kurdistan, 1991*
Northern Somalia, 1991*
Croatia, 1991
Bosnia, 1992
Abkhazia, 1992-93'
Eritrea, 1993
Chechnya, 1994"

Poland, 1918
Ireland, 1921
India, 1947
Palestine, 1947
Cyprus, 1974"

*De facto, not internationally recognized.


Cases of decolonization in which the colony and its inhabitants had never been part of the
metropolitan state are not included, because in most instances the remaining metropolitan
population was small and viewed by most inhabitants as foreigners, not as a local ethnic
group that could potentially contest for power. Algeria is an exception, because it had
been legally part of the French metropole and also contained an ethnic French population
of more than 1 million who wanted both t o stay where they were and t o remain French
citizens.
No large-scale ethnic violence within or among the successor states, although one successor state (Hungary) fought a war with an existing state (Romania) that had gained territory
in the partition.
No wars among successor states, but Turkey fought a war with Greece in 1920-23.

NOTES:
a

IRELAND

Although the partition of Ireland has been accompanied by violence, this


violence was not caused by partition itself, but by the fact that the partition
did not separate the antagonistic communities, particularly in the North. As a
result, the demographically mixed North has been subjected to decades of
violence, while the relatively monolithic South has enjoyed peace.
DID THE PARTITION OF IRELAND REDUCE VIOLENCE OR INCREASE IT? Political
violence in Ireland increased markedly in the decade after the British government agreed to Home Rule and to partition in 1914. There were four major
episodes: (1) the Easter Rising of 1916, in which approximately 450 died;14
14. Alan J. Ward, The Eustev Rising (Arlington Heights, IU.:AHM Publishing, 1980), p. 13.

When All Else Fails I 127

(2) the Irish War of Independence from 1919 to 1921, in which an estimated
1,500 people were killed;15 (3) the 1922-23 civil war within the Irish Republic,
which cost as many as 4,000 more lives;I6 and (4) sectarian violence in Ulster
between 1920 and 1922, which left another 428 dead.I7
Opponents of partition and separation might see this record as evidence for
their position, but this would be a mistake. The first two episodes were not
sectarian conflicts at all and had nothing to do with partition; rather, they were
fought over whether Ireland would receive only Home Rule within the United
Kingdom or whether it would gain full independence. The opposing sides in
both rounds were Irish nationalists versus the British Army and police, not
Catholics versus Protestants.
The third and fourth episodes occurred because the 1921 partition did not
separate the two communities, particularly in Ulster. Separation was not an
issue in the South, because the twenty-six counties assigned to the Irish Free
State contained a population that was less than 10 percent Protestant.8 This
minority was far too small and too thinly spread to constitute a possible
political or military force. Thus there was no security dilemma, and Ireland
has not experienced a problem of sectarian violence to this day.
The source of both the sectarian violence in Ulster and the Irish Civil War
was the mixed demography of Northern Ireland. The six counties that remained part of Britain included a 34 percent Catholic minority, creating a fairly
intense security dilemma. With extreme nationalists in the South calling for
action to undermine the partition, many Ulster Protestants believed that preserving their political, economic, and possibly even physical security required
suppressing any accretion of power by Catholics. This security dilemma was
further exacerbated by the irregular and commingled settlement patterns of
the two groups within Northern Ireland (see Map 1);if Catholics and Protestants had been mostly separated in distinct regions, then Protestants would
still have controlled the province government, but would have had little reason
to interfere in local rule of Catholic areas. The result in 1920-22 was a wave of
15. Michael Hughes, Ireland Divided: The Roofs of the Modern Irish Problem (New York St. Martins
Press, 1994), p. 49; and Ward, Easter Rising, p. 126.
16. J.J. Lee, Ireland, 2922-1 985: Politics and Society (Cambridge, U.K.: Cambridge University Press,
1989), p. 69.
17. Patrick Buckland, Ulster Unionism and the Origins of Northern Ireland (Dublin: Gill and Macmillan, 1973), p. 176; and Charles Townshend, The British Campaign in Ireland, 1919-1921 (London:
Oxford University Press, 1975), p. 342.
18. 1911 Census, reported in Buckland, Ulster Unionism, pp. 179-180.
19. Ibid.

International Security 23:2 1 128

both organized and private violence in which 428 people were killed, more
than 8,000 Catholics were forced out of their jobs, and about 23,000 were driven
from their homes.20
Northern Ireland's intermixed demography also created a security dilemma
for nationalists in the South, who saw the vulnerability of the Ulster Catholics
as demanding action to rescue them. This security dilemma did not lead to
international war (the Irish government recognized that undoing the partition
by force was infeasible), but it did help cause a short civil war in the South.
The Irish Republican Army (IRA), which was operating in the North, refused
to accept government discipline and was defeated by government forces in a
war lasting from June 1922 to August 1923.21One could argue that even if there
had been virtually no Catholics resident in the North, the IRA would have
fought anyway for the cause of a united Ireland, but IRA leaders' appeals to
their membership to die for mere land would have been much less compelling
than a call to rescue fellow Catholics from pillage and murder. Similarly, it
would have been harder to accuse the government of treason for abandoning
people who were not there.22
COULD THE PARTITION OF IRELAND HAVE BEEN AVOIDED? The partition of
Ireland could have been averted in only two ways, either of which would have
had worse consequences.
First, Britain could have granted Home Rule (or independence) to a united
Ireland and coerced the Ulster Unionists into submission. In fact, the British
government tried to do this in 1914, but was stopped by the risk of civil war
in Britain itself, as well as by the evident willingness and capability of the
Unionists to resist. Imposing Home Rule on Ulster was opposed by the Tory
opposition in Commons, a majority of the House of Lords, King George V, and
a great many Army and Navy officers.23

20. Buckland, Ulster Unionism, p. 176; D.G. Pringle, One Island, Two Nations? (Letchworth, U.K.:
Research Studies Press, 1985), pp. 239-242; and Frank Gallagher, The Indivisible Island (London:
Victor Gollancz, 1957), pp. 225-265.
21. J. Bowyer Bell, The Secret Army, 2916-1970 (New York John Day, 1971), pp. 29-66; and Dennis
Kennedy, The Widening Gulf: Northern Attitudes to the Independent Irish State, 192949 (Belfast:
Blackstaff Press, 1988), pp. 72-77.
22. In the June 1922 election, the most extreme nationalist party, Eamon De Valera's faction of Sinn
Fein, gained only 36 seats versus 92 others, but even De Valera did not favor prompt action against
Northern Ireland. The IRA was not united either; an emergency convention in June voted narrowly
against rebellion. Bell, Secret Army, pp. 3C34.
23. When in March 1914 the British government attempted to deploy troops to reinforce arms
depots in Ulster, most of the officers who received the orders mutinied. War Minister J.E.B. Seely,
Army Chief of Staff Sir Henry Wilson, and Army Commander-in-ChiefSir John French sided with

When All Else Fails I 129

Map 1. Religious Majorities in Ulster, 1911.

0Catholic
SOURCE:

Office).

Protestant

Data taken from the Census of Ireland, 7977 (London: His Majesty's Stationery

International Security 23:2 I 130

In addition, few if any doubted that the Ulster Protestants would fight.24On
Ulster Day, September 18,1912,218,206 Ulstermen signed a Solemn League
and Covenant, pledging to resist Home Rule; 228,991 women signed a parallel
document. This was most of the adult Protestants in the province.25Since 1912
they had been preparing a regional defense force, the Ulster Volunteer Force,
that by 1914 had 85,000 to 90,000 members, and was fairly well organized and
rapidly improving its equipment through purchase and smuggling.26It was
said in London in the spring of 1914 that suppressing Ulster would have
required the entire British Army and would have taken twelve to eighteen
months.27
The other possibility would have been to deny Irish independence indefinitely. This would have required suppressing all armed combinations and
nationalist political cells throughout Ireland, a task that 40,000 British troops
and 10,000 police could not carry out in 1919-21. The British would certainly
have been forced to engage in brutal tactics against civilians, an alternative
that the government rejected as too
DOES PARTITION INCREASE HATRED AND GENERATE NEW CONFLICTS? In the
seventy-five years since partition, the Irish Free State has enjoyed virtually
complete freedom from sectarian conflict. A united Ireland remains a rhetorical
goal, but public support for action toward this end has faded over the years.
Even though the 1985 Anglo-Irish agreement on Northern Ireland contained
no promise of eventual sovereignty, the Irish government paid no domestic
political price for signing
the mutineers, assuring them that force would not be used against Ulster. When ordered to
withdraw this promise, Seely and French resigned, which led to a threat of mass resignations by
military officers. At this point there was a real risk of a complete split in the British Army. Elizabeth
A. Muenger, The British Military Dilemma in Ireland: Occupation Politics, 2886-1914 (Lawrence:
University Press of Kansas, 1991), pp. 168-172, 188-191; and Hughes, Ireland Divided, pp. 34-36.
24. Buckland, Ulster Unionism, p. 64;Townshend, Political Violence, p. 343; and Gallagher, The
lndivisible Island.
25. John F. Galliher and Jerry L. DeGregory, Violence in Northern Ireland: Understanding Protestant
Perspectives (New York Holmes and Meier, 1985), p. 10. Despite class, rural-city, and other cleavages within the Ulster Protestant community, Unionism commanded absolutely solid support. In
1910 all Protestant-majority districts in Ulster elected Unionist MIS. Buckland, Ulster Unionism,
frontispiece, pp. 22-34, 179-180.
26. Townshend, Political Violence, pp. 252-255; and Muenger, British Military Dilemma, p. 177.
27. Hughes, Ireland Divided, p. 35.
28. Townshend, British Campaign in Ireland, pp. 189-192; and Sheila Lawlor, Britain and Ireland,
1914-2923 (Dublin: Gill and Macmillan, 19831, pp. 8586.
29. In 1991,82 percent were prepared to postpone union if that would help bring about an internal
settlement in the North. Market Research Bureau of Ireland, reported in Gemma Hussey, Ireland

When All Else Fails 1 232

Northern Ireland, however, has not achieved peace between the two religious communities. Protestants still fear eventual Catholic rule (or union with
Ireland), Catholics have been subjected to decades of discrimination, and
repeated rounds of political violence have erupted.30The Troubles from 1969
to the present have claimed about 3,400 lives.31The sources of the difference
between these two histories are again demography and the security dilemma.
The Protestant population of Ireland stood at about 8.5 percent in 1991, while
the Catholic minority in the North had increased to 38 per~ent.~
Resolution of the conflict in Northern Ireland requires assuring the security
of both communities, which in turn requires a credible, joint guarantee by the
only actors strong enough to provide it-the Irish and British governments.
Very recently, prospects for peace have improved because the two governments finally seem prepared to provide whatever assurances are necessary. In
April 1998 the Irish Dail formally amended the republics constitutional claim
on the North to require the consent of the provinces population, and in May
the Irish government condoned British Prime Minister Tony Blairs promises
to Unionists that violent factions would be excluded from the new Northern
Ireland Assembly and that there would be no change in the status of Northern
Ireland without the peoples consent.%
WHAT SHOULD HAVE BEEN DONE? The partition of Ireland was unavoidable,
but forcing 430,000 Catholics into Northern Ireland was not. Unfortunately, the
Ulster Unionists commanded so much political support in Britain at the time
that they were in effect allowed to set the boundaries of Northern Ireland.
Seeking to incorporate as many Protestants as possible without endangering
their majority in the province, they chose six of Ulsters nine counties, including two that contained more Catholics than protest ant^.^^

Today (London: Viking, 1993), pp. 186-188. In a 1996 poll only 38 percent of people polled in the
republic supported unification. Carl Homore, Desire for Union Now a Need for Peace, Houston
Chronicle, September 22, 1996.
30. Richard W. Mansbach, ed., Northern Ireland: Halfa Century of Partition (New York Facts on File,
1973).
31. Joseph Ruane and Jennifer Todd, The Dynamics of Conflict in Northern Ireland: Power, Conflict,
and Emancipation (Cambridge, U.K.: Cambridge University Press, 19961, p. 1.
32. Ireland: Statistical Abstract, 2995 (Dublin: Stationery Office, 1995), pp. 55, 409.
33. Dublin Parliament Poised for Peace Poll Go-Ahead, Press Association, April 21, 1998; and
Frank Millar, Blair and Trimble Appeal to Undecided Unionists, Irish Times, May 21, 1998.
34. Lawlor, Britain and Ireland, 1914-2923, pp. 124-126; and D.W. Harkness, Ireland in the Twentieth
Century: Divided Island (New York St. Martins Press, 19961, pp. 3p37.

International Security 23:2 132

Irelands best chance at lasting peace would have been to draw a partition
line that separated the two groups as fully as possible, without regard to
county lines or other prior b ~ u n d a r i e sBecause
.~~
no line could avoid leaving
substantial minorities on each side, the British government should also have
offered money to people willing to move, making clear that it could not protect
those who insisted on staying behind. The result would have been a smaller
but safer Northern Ireland.
INDIA AND PAKISTAN

The most frequently mentioned case in the debate over ethnic population
transfers and partitions is India. Critics of the 1947 partition blame it for
causing more than 15 million refugees and hundreds of thousands of deaths.
This correlation, however, is spurious. The partition, the population transfers,
and the violence were all caused by the irresolvable security dilemmas between the Muslim and Hindu communities of India, and especially between
the Muslim and Sikh communities of Punjab Province, both of which were
generated by the removal of the imperial power that had previously guaranteed the security of all groups. In short, independence from Britain, not partition, caused these tragedies.
Constructivist scholars of identity would charge the communalization of
Indian politics, and therefore ultimately the violence, to manipulation of mass
aspirations and fears by self-interested communal elites.36Whether or not we
accept this model of mass political mobilization, however, the fact remains that
the removal of British imperial power created real security dilemmas, which
lent inherent credibility to political appeals based on community security.
DID PARTITION CAUSE THE VIOLENCE? The independence of India and Pakistan on August 15, 1947, not only divided the Indian subcontinent; it also
involved the partition of two of colonial Indias most populous provinces,
35. A Boundary Commission operated from 1924 to 1925, but was limited to recommending only
very minor changes, which in any case were not implemented. Report of the Irish Boundary
Commission, 1925 (Shannon: Irish University Press, 1969).A better partition would have given the
Unionists all of counties Antrim and Londondeny; much but not all of Down, Armagh, and Tyrone;
and a few small bits of Monaghan, Fermanagh, and Donegal. The large Catholic-majority region
spanning Londondeny and Tyrone (see Map 1) is mountainous and was thinly settled.
36. Paul R. Brass, Language, Religion, and Politics in North India (London: Cambridge University
Press, 1974); Mushirul Hasan, Nationalism and Communal PoIitics in India, 1916-1 928 (Columbia,
Mo.: South Asia Books, 1979);Gyanendra Pandey, The Construction of Communalism in Colonial North
India ( D e w Oxford University Press, 1990); and Milton Israel, Communications and Power: Propaganda and the Press in the Indian Nationalist Struggle, 1920-1947 (Cambridge, U.K.: Cambridge
University Press, 1994).

When All Else Fails 133

Punjab in the northwest and Bengal in the east. Although intercommunal


rioting had been on the rise in 1945 and 1946, independence and partition were
expected to dampen the violence. Some population movements were anticipated, but they were not expected to be especially large, sudden, or dangerous.
These expectations were met in Bengal, where more than 5,000 people were
killed before independence but very few after, and where 3.5 million people
moved across the new border with little loss of life.
Punjab accounted for most of the refugees and nearly all the deaths. From
August to October 1947, the province was convulsed by an intense communal
civil war involving some of the largest ethnic cleansing campaigns in history.
Hundreds of thousands of people were killed in Punjab, and the war sparked
large numbers of revenge killings elsewhere as well. More than 10 million
people from the Punjab and adjacent provinces had to flee for their lives.
By the late 1920s it was clear to all parties that India would achieve eventual
independence. Given that British power was the ultimate guarantee of security
for all communities in India, the prospect of its withdrawal activated potential
intercommunal security dilemmas. Two such dilemmas were critical in determining the final outcome of the process: between Muslims and Hindus at the
national level, and between Sikhs and Muslims in P ~ n j a b . ~ ~
Muslims made up 22 percent of the population of India and Hindus 68
percent?8 meaning that under pure majoritarian rule the Muslims would be
absolutely insecure in the event that the government should be captured by
Hindu supremacists such as the Hindu Mahasabha movement. Although the
largest Indian nationalist movement, the Congress Party, was formally committed to a secular India, in practice it never represented all Indian communities. Members of the Mahasabha movement and other Hindu nationalists, such
as B.S. Moonje, were welcome in Congress, while members of Muslim parties
were excluded as c~mmunalist.~~

37. The Hindu-Muslim security dilemma was most severe in the belt of North India, where the
percentage of Muslims ranged from 20 to 60 percent. In the South, where Muslim minorities were
quite small, intergroup security dilemmas were weak, and communal mobilization and violence
remained low before, during, and after partition.
38. Census of India, 1931, Vol. 1, Part 1 (Delhi: Manager of Publications, 1933), p. 392.
39. H.V. Hodson, The Great Divide: Britain-lndia-Pakistan (Oxford, U.K.: Oxford University Press,
1985), p. 59. Moonje argued that Hindus should negotiate with Muslims only from a position of
strength and that Gandhis philosophy of nonviolence would lead to destruction and extermination of the Hindus from the face of the world. Letter, March 16, 1922, cited in Mushirul Hasan,
Communalist and Revivalist Trends in Congress, in Hasan, ed., Communal and Pan-Islamic Trends

International Security 23:2 I 234

Mohammad Ali Jinnah, leader of the Muslim League, demanded constitutional arrangements that would assure communal autonomy, especially: (1
guaranteed electoral majorities in the five provinces with Muslim-majority
populations-Punjab, Bengal, Northwest Frontier Province (NWFP), Sind, and
Baluchistan; and (2) a weak federal system in which the central government
would have little power over the provinces?' Muslim leaders further insisted
that police functions should belong to the provinces and national defense to
the British governor-general (i.e., that a Hindu-dominated central government
should possess no tools of coercive force).41Agreement in principle on some
of these points was reached at an All-Parties Conference in 1928, but the
Congress Party leadership was forced to disavow the deal because of Hindu
nationalist opposition.42In practice, the Muslim demands might have exposed
Hindu minorities in Punjab, Bengal, and Sind to Muslim domination.
The results of the 1937 elections intensified this security dilemma in three
ways. First, electoral success persuaded Congress leaders that they could reach
out directly to the Muslim masses, ignoring the Muslim political parties.43To
survive, the Muslim League had to transform itself from an elite circle into a
genuine mass party that could claim to represent most Indian Muslims. It
succeeded, based on explicitly communal appeals such as the slogan "Islam in
Danger," so the ultimate effect was to increase fear and mistrust between the

in Colonial India (Delhi: Manohar, 19851, p. 206. According to JawarlhalalNehru, many a congressman "was a communalist under his national cloak." Nehru, An Autobiogruphy (London:John Lane,
1936), p. 136.
40. Jinnah also called for onethird of the seats in the central legislature and a 75 percent majority
requirement for action by the legislature. Although Jinnah did not dominate Muslim politics until
much later, a wider meeting of a number of Muslim groups in 1927 had agreed on a similar
program, as had another such meeting in 1925. V.P. Menon, The Transfer of Power in India (Princeton,
N.J.: Princeton University Press, 1957), pp. 36-37; Uma Kaura, Muslims and Indian Nationalism
(Columbia, Mo.: South Asia Books, 19771, pp. 29-30; and R.J. Moore, The Crisis of Indian Unity,
2927-1940 (Oxford, U.K.: Clarendon Press, 1974), pp. 24-25.
41. Aga Khan at the Second Round Table Conference, London, 1931, reported in Kaura, Muslims
and Indian Nationalism, pp. 72-73.
42. Moore, Crisis of Indian Unity, pp. 101-104; Uma Kaura, Muslims and Indian Nationalism, pp. 4251; R. Coupland, The Indian Problem: Rcport on the Constitutional Problem in India, Vol. 2 (New York
Oxford University Press, 19441, p. 125; and Hasan, "Communalist and Revivalist Trends in Congress,'' p. 210.
43. Congress swept most of the country, gaining control of seven provinces. Subsequently, it took
the position that it represented all Indian nationalists, so its own Muslim members, not Jinnah or
others, were the true arbiters of Muslim opinion and interests. Actually, however, it contested only
58 of 482 Muslim seats and won just 26,19 of these based on Pathan protest votes against Punjabi
political dominance of Northwest Frontier Province. Hodson, The Great Divide, pp. 66-72; Return
Showing the Results of the Elections of 1937, Cmd. 5589, cited in Coupland, Indian Problem, pp. 1.516,
121-123.

When All Else Fails I 135

two communities. Second, Muslims in the seven provinces ruled by Congress


soon complained of abuses, including physical insecurity because of government failure to restrict communal violence by Hindus; these reports were
widely circulated.@Third, Congresss aggressive political tactics led M u s h
leaders to doubt whether British control over defense could last much beyond
independence. A Congress-controlled government might contrive to reduce the
governor-general to a figurehead as it had in Australia, then change the basis
of recruiting of the army, and then be able to do anything at all.45
By 1940 Jinnah was convinced that nothing short of a separate Muslim state,
Pakistan, could provide security for the Muslim community. Further HinduMuslim negotiations proved fruitless; and in the face of increasing Muslim
unity and evident determination, on June 3, 1947, Congress, the Muslim
League, and Viceroy Earl Louis Mountbatten agreed to partition.& It was
further agreed that because both Punjab and Bengal contained only narrow
Muslim majorities and large Hindu-majority regions, Muslim West Punjab and
East Bengal would go to Palustan and predominantly Hindu East Punjab and
West Bengal to India. Both states would gain independence on August 15,1947,
and the British-chaired Boundary Commissions would announce their decisions on August 17!7
Although the national-level Hindu-Muslim security dilemma necessitated
the partition of India, it did not cause most of the tragedy that followed, except
indirectly by distracting the major players attention from a second, even more
severe, security dilemma between Muslims and Sikhs in the province of Punjab. Virtually all of the Sikh population of nearly 6 million in 1941 was concentrated in Punjab. Although the provincial population overall was 56 percent
Muslim, 27 percent Hindu, and just 13 percent Sikh, the Sikhs averaged
considerably wealthier than the other communities and had exercised disproportionate power in provincial politics.@By the 1940s Sikhs and Muslims had
44. Rajendra Prasad, India Divided (Bombay: Hind Kitabs, 1946), pp. 146-152; and Stanley Wolpert,
Jinnah of Pakistan (New York: Oxford University Press, 1984), pp. 164-169.
45. Kaura, Muslims and Indian Nationalism, pp. 128-129.
46. In 1946 elections the Muslim League won all the Muslim seats in the central assembly. In the
Punjab legislature, the Muslim League won 79 of 86 Muslim seats, while the intercommunal Punjab
Unionist Party declined from 99 seats in 1937 to 18. Two of these immediately defected to the
Muslim League. E.W.R. Lumby, The Transfer of Power in India, 1942-7 (London: George Allen and
Unwin, 1954), pp. 69, 145-148.
47. Ibid., pp. 162-164.
48. Census of India, 1931, Vol. 1, Part 1, pp. 387,392; and Census of India, 1942 (Delhi: Manager of
Publications, 1943). Prior to 1927 Sikhs were 24 percent of the electorate and held 18 percent of

International Security 23:2 236

not fought in several decades, but they had a long history of intercommunal
h0stility.4~Sikhs also controlled a large fraction of the best land in Central
Punjab and in the canal colonies in West Punjab (see Map 2). Accordingly, Sikhs
feared Muslim dominance even more than Muslims feared Hindu rule; their
wealth, political influence, religious freedom, and even physical security all
might be at risk. From the Muslim point of view, the Sikhs presented a special
threat because their martial tradition meant that the whole male population
had to be considered armed.50
The core Sikh concerns were driven by security and remained essentially
unchanged from the 1920s onward. In different negotiations with various
parties, Sikh demands varied considerably on issues such as representation in
the Punjab legislature, the boundaries of Punjab, and representation in the
central legislature; but they remained constant on the two points most central
to their physical security: (1) creation of a political unit in which Sikhs would
be, if not a majority, at least holders of the political balance between Muslims
and Hindus (i.e., anything but a Muslim-majority Punjab); and (2) retention of
their traditional overrepresentation in the army of whatever state they would
be part
To secure these goals, Sikh leaders pursued every possible avenue of negotiation: first, participation in all-India constitutional negotiations; then, direct
negotiations with Congress. After it became clear that partition could not be
avoided, they tried direct negotiations with the Muslim League on Sikh status
within Pakistan, made an attempt to get the British to impose arrangements
for Sikh security, offered a proposal for an independent Sikh state, and finally

the seats in the provincial council. Sikhs also maintained special claims to Punjab on grounds that
they provided a disproportionate share of the province's revenue, that the province contained all
their important religious sites, and that Sikhs had been the last rulers of the region before the
British. Anup Chand Kapur, The Punjab Crisis (New Delhi: S. Chand, 1985), pp. 39, 43.
49. Sikhism was transformed from a purely religious movement to a distinct, militant community
in response to Muslim rule; Sikhs fought 250 years of almost continuous war against Muslim
princes from about 1600 onward. Sikh rulers had controlled the Punjab for about 100 years prior
to the British takeover in 1849, and were still remembered in the twentieth century for harsh
repression of their Muslim subjects. Sikh troops also helped suppress Muslim rebels during the
Sepoy Mutiny in 1857-58. Kapur, Punjab Crisis, pp. 6-9; and Hodson, Great Divide, pp. 1&20.
50. Sikh tradition requires males to carry ceremonial daggers at all times, and to take up arms
when necessary to defend righteousness. Sikhs also comprised a greatly disproportionate fraction
of the Indian Army: 13 percent in 1930 compared to slightly more than 1 percent of the population.
Kapur, Punjab Crisis, p. 7, notes 9 and 10; and p. 20, n. 47.
51. Ibid., pp. 51, 98, 111.

Muslim and Sikh Populations in Punjab, 1941.

1
Sikh

uslirn

ata taken from the Census of India, 7947 (New Delhi: Manager of Publications, 1943).

>65%

50-65%
35-50%

20-35%

10-20%
0-1 0%

International Security 23:2

I 138

put forth a proposal to partition Punjab along lines that would keep all Sikhs
within India. All were rejected.52
By the summer of 1947 Sikh leaders were desperate. On March 2 the Punjabs
coalition government made up of the Congress Party, Sikhs, and the crosscommunal Unionist Party had collapsed in the face of a massive Muslim
League civil disobedience campaign.53In February and March Muslims had
attacked Hindus and Sikhs in Lahore and Amritsar, the two main cities of
central Punjab and the core of the region most likely to be disputed; more than
3,000 were killed.54Possibly most threatening had been Jinnah's proposal in
December 1946 to Sikh leader Sardar Baldev Singh that the Muslims and Sikhs
combine to seize all of Punjab, while still refusing to provide any guarantees
of the status of Sikhs in Pakistan. This could only inflame key Sikh fears that
Muslim rule would be oppressive, and that the Muslims would not be satisfied
with any initial territorial ~ e t t l e m e n tFinally,
.~~
it was clear that the June 3
partition agreement would, in all likelihood, leave nearly 2 million Sikhs
stranded in Pakistan.
It appears that at this point Sikh leaders devised a four-point last-resort plan
to protect their national security unilaterally: (1) if the boundary award proved
unsatisfactory, to contest as much as possible of the core Sikh areas in central
Punjab, and to resist possible Muslim attempts to contest any part of East
Punjab; (2) to evacuate most Sikhs west of the line; (3) to eliminate the Muslim
population east of the line, thus increasing the Sikh percentage in East Punjab

52. Ibid.; Satya M. Rai, Partition of the Punjab (New York Asia Publishing House, 1965),pp. 37-38;
and Lumby, Transfer of Power, 1942-7, pp. 185-186.
53. This agitation was sparked by a government attempt in January to disarm the Muslim National
Guards. Ian Talbot, Khizr Tiwana, the Punjab Unionist Party, and the Partition of India (Surrey, U.K.:
Curzon Press, 1996), pp. 68, 148, 154-161.
54. This violence was partly sparked by S i leader Master Tara Singhs calls in the provincial
assembly on March 4 for "Death to Pakistan" and "The pure shall rule; no resister will remain."
Penderel Moon, Divide and Quit: A n Eyewitness Account of the Partition of India (London: Chatto and
Windus, 19621, p. 77. The police in Lahore and Amritsar, who were mostly Muslim, were not
effective. Violence also spread to Muslim-majority areas in Western Punjab, in turn followed by
attacks on Muslims in Southeastern Punjab. Ian Talbot, Punjab and the Ruj (Riverdale, Md.: Riverdale, 1988), pp. 227-228; h i , Partition of the Punjab, p. 83, n. 25; and Anita Inder Singh, The Origins
of the Partition of India, 1936-1947 ( D e w Oxford University Press, 1987), pp. 218-220.
55. Singh, Origins of Partition, pp. 205206; and Lumby, Transfer of Power, 1942-7, p. 186. In June a
last attempt by moderate Muslim League leader Nawab Mushtaq Achmad Gurmani to mediate
between the Muslim League and the Sikh leadership was disavowed by Jinnah. Moon, Divide and
Quit, pp. 82-87. According to Moon, Jinnah had frequently threatened the Sikhs that division of
Punjab would not be in their interest, but had never encouraged them on their prospects in
Pakistan. See also Kapur, Punjab Crisis, pp. 94-95.

When All Else Fails 239

after the war; and (4) later, to press the Indian government for redivision of
East Punjab in order to create a true Sikh-majority province.
Although the existence of such a plan cannot be established with certainty,
there is suggestive evidence. As early as March 1947, the Sikh Panthic Party
passed a resolution that it would fight Pakistan to the end.56Military mobilization began in April, and by June the Sikh Akali Fauj had 8,000 men; in
addition, the British provincial governor was receiving intelligence reports of
a plan for a terror campaign in East P ~ n j a bSeveral
. ~ ~ Sikh leaders, including
Sardar Baldev Singh, the Sikh representative on the Boundary Commission,
made clear that they would not respect an unfavorable award.58Fighting began
well before the award was even announced; starting on July 30, Sikh forces
attacked Muslim villages in the central region around Lahore and Amritsar,
which was disputed by both communities, as well as Muslim communities
throughout East P ~ n j a bBy
. ~the
~ end of August, much of East Punjab had been
cleared of its Muslim population.60
The main evidence that the Sikh evacuation from West Punjab was preplanned is that most Sikhs departed before trouble broke out in their areas;
they also demonstrated better preparation than Muslim refugees going the
other way and succeeded in getting away with more of their property and
fewer losses.61After the war Sikhs numbered roughly 30 percent of the Indian
province of East Punjab and Hindus 70 percent. Sikhs then agitated for further
division of Punjab to create a Sikh-majority state, which was accomplished in
1966.62
('

56. This was the only organized Sikh party, holding 22 of 33 Sikh seats in the Punjab assembly.
Talbot, Punjab and the Raj, p. 227; Rai, Partition of Punjab, p. 40; and Kapur, Punjab Crisis, p. 50.
57. By June the Hindu RSSS had 58,000 men and the Muslim League National Guards 39,000.
Talbot, Punjab and the Raj, pp. 232-233. At the end of March the superintendent of police in Delhi
predicted that "once a line of division is drawn in the Punjab all Sikhs to the West of it and all
Muslims to the east of it will have their...chopped off." Quoted in Moon, Divide und Quit,
pp. 87-88. The previous governor of Punjab had begun predicting civil war as early as 1945.
Wolpert, Jinnah, p. 249.
58. Talbot, Punjab and the Raj, p. 232; and Hodson, Great Divide, p. 338.
59. Although in the end this fighting did not affect the final partition line, it appears that Sikh
forces did temporarily occupy some locations west of the partition line. Talbot, Punjab and the Raj,
pp. 233-234; The Sikhs in Action (Lahore: Superintendent of Government Printing, 1948);and Moon,
Divide and Quit, pp. 151-152.
60. Moon's judgment is that the Sikhs were deliberately making mom for 2 million refugees from
Pakistan. Divide and Quit, pp. 279-280.
61. Hodson, Great Divide, p. 411; and Moon, Divide and Quit, pp. 122, 281.
62. Kapur, Punjab Crisis, pp. 149-183. This did not end the conflict over the scope of Sikh autonomy, which flared into violence again between 1980 and 1992.

International Security 23:2 140

The war in Punjab also sparked additional ethnic cleansing in adjacent


provinces. By September refugees bearing tales of atrocities flooded Delhi and
the United Provinces, leading to revenge massacres of Muslims, which in turn
led to Muslim attacks on Hindus and Sikhs in West Pakistani cities such as
Peshawar and Karachi. Both sides also attacked refugee convoys and trains
passing through Punjab itself.63Altogether, hundreds of thousands of people
were killed, and more than 10 million refugees were exchanged between India
and West Pakistan.@
An additional cause of loss of life in this war was the failure of the British,
Indian, or Pakistani governments to take meaningful preparations to protect
refugees in Punjab, in part because they all underestimated the scale and
suddenness with which the war would escalate, but also because the Indians
and Pakistanis did not want to legitimate or encourage population movem e n t ~The
.~~
British-commanded Indian Boundary Force, which was supposed
to control communal violence in Punjab, was, at 50,000 men, far too small for
the task, and many of its contingents proved unreliable for communal reasons.
Transport and reception camps were not prepared. For weeks, the Indian
government also continued to send Muslim refugee trains directly through
Punjab, rather than around it.66
At no point did any side make energetic efforts to protect, feed, or shelter
refugees in transit through Punjab. Many lives lost in Punjab could have been
saved if refugees had not been directed toward the center of the Sikh uprising.
Nearly all could have been saved had the British provided enough reliable
troops to control the province, or found allies who could. The lesson of Punjab
is not that population movements must be costly, but rather that refugees
should not be forced to travel through a war zone.67
63. In many areas police and officials of the "wrong" community fled, further reducing restraints
on pogroms. Menon, Transfer of Power, pp. 419423; Hodson, Great Divide, p. 406; and Lumby,
Transfer of Power, 1942-7, pp. 193-195.
64. The number of deaths is disputed. Hodson estimates 200,000, Moon less than 200,000. Lumby
says "hundreds of thousands," an Indian High Court judge later estimated 500,000, and the
Pakistani government claimed more than 500,000 Muslims alone. Kumar says that more than a
million died. Hodson, Great Divide, p. 418; Moon, Divide and Quit, pp. 268-269,293; Lumby, Transfer
of Power, 1942-7, p. 199; Sikhs in Action, foreword; and Kumar, "Troubled History," p. 26.
65. Congress agreed to "communal option" for officials, but maintained that the general population throughout the country should stay where they were. Rai, Partition of the Punjab, pp. 73-75.
66. Hodson, Great Divide, p. 412; Rai, Partition of the Punjab, p. 79; and Moon, Divide and Quit,
pp. 278-279.
67. Barry Posen provides a formula that can be used to estimate the number of troops needed to
protect a refugee movement, in this case roughly 250,000. Posen, "Military Responses to Refugee
Flows," International Security, Vol. 21, No. 1 (Summer 19961, p. 106, n. 51.

When All Else Fails I 142

Although the war in Punjab accounted for the vast majority of all the deaths
in communal conflict between 1945 and 1947, one other region of India is also
especially important to this analysis. Unlike Punjab, the partition of Bengal
markedly dampened the security dilemma in that province; it assured that the
millions of Hindus in West Bengal would not have to live under Muslim rule,
and, unlike Punjab, the division line did successfully separate districts populated mainly by each of the two communities. Most important, there was no
third side with an overwhelming security motive to overturn the settlement.
As a result, the announcement of independence and of the partition line
lowered violence in Bengal by resolving both sides' security uncertainties.
More than 5,000 were killed in the province in the year before independence,
but very few afterward.68Between 1947 and 1951, 3.5 million people moved
between India and East Pakistan in orderly, planned transfers, without loss of
life.
DID PARTITION INCREASE HATRED AND CAUSE NEW CONFLICTS? The 1947-51
population exchanges resolved Hindu-Muslim security dilemmas throughout
most of India and Pakistan, as very few Hindus remained in Pakistan, while
the Muslims of India are too few, too thinly spread, and too far from any
possible aid to even imagine resisting the Indian government . . . and they
have not. Sporadic Hindu-Muslim violence still occurs in India, although at
low levels compared with the fears of both sides from the 1920s to the 1940s
or the actuality of 194547.
The independence of India and Pakistan did generate one new conflict, over
control of the former princely state of Jammu and Kashmir. This conflict
occurred not because India was partitioned but because Kashmir, whose population was about two-thirds Muslim, was not. Drawing a partition line through
Kashmir would have been easier than in Punjab or even Bengal, because most
of the Hindu population of the state resided in the southernmost division,
Jammu, adjacent to India, and the boundary between Jammu and the rest of
the state is largely m o u n t a i n ~ u s . ~ ~

68. Although accounts are incomplete, it appears that total deaths in intercommunal violence in
India in 1946-47 not directly related to the Punjab civil war may have been about 20,000. Suranjan
Das, Communal Riots in Bengal, 2905-1947 (Delhi: Oxford University Press, 1993), pp. 167-205;
Francis Tuker, While Memory Serves (London: Cassel, 1950), pp. 424-426; Keesing's Contemporary
Archives, January 17-24, 1948, p. 9049; Menon, Transfer of Power, pp. 294, 434-445; and Lumby,
Trunsfer of Power, 2942-7, pp. 120-122.
69. Majid Husain, Geography of Jammu and Kashrnir State (New Delhi Rajesh Publications, 1987),
p. 54.

International Security 23:2 I 142

Failure to divide Kashmir intensified Hindu-Muslim security dilemmas in


three ways, generating a history of conflict that still continues. First, although
communal relations in Kashmir had been better than in many other areas, and
Maharaja Hari Singh initially attempted to keep Kashmir independent, by
October 1947 each community had reason to fear for its security. Both groups
were aware that India and Pakistan each claimed Kashmir, both had heard
about atrocities in Punjab, and some of the maharaja's troops began attacking
the Muslim p~pulation.~'
Second, Kashmir borders key economic centers of both countries, and is thus
strategically valuable. In October the maharaja invited in pro-Indian Sikh
troops, and a few weeks later Muslim irregulars invaded from Pakistan. Both
regular armies intervened. Battle deaths reached 1,500 before a truce was
signed at the start of 1949.71India took control of most of the state, but Pakistan
invaded again in 1965, and there was more border fighting in 1971.
Third, under Indian rule Kashmiri politics have become increasingly communalized over time, threatening the security of all groups. From the early
1960s onward, increased political participation has led to several cycles of
Muslim autonomy demands and Indian government responses that actually
reduced local authority. For example, from 1980 to 1982 the Kashmir government backed a proposal to allow 1947 refugees (i.e., Muslims) to return,
sparking fears that Hindus and S k h s now settled on refugees' former property
would be dispossessed. When the 1983 election again returned the same
government in a vote along communal lines, Indian President Indira Gandhi
removed the state government and instituted repressive measure^.^' Since the
late 1980s Kashmir has been fighting an ongoing Muslim insurgency, aided by
Pakistan. More than 30,000 people have been killed, and virtually the whole
Hindu population of the Valley of Kashmir (about 250,000) have fled their
homes. It is uncertain how many Kashmiri Muslims support the insurgents,
but nearly all have become profoundly alienated from Indian government
Since May 1998 Kashmir has become the focus of mutual Indian and
Pakistani nuclear threats.
70. Sumit Ganguly, 7'he Origins of War in South Asia: Indo-Pakistani Conflicts since 1947 (Boulder,
Colo.: Westview Press, 1994), pp. 34-35, 42.
71. Alastair Lamb, Kashmir: A Disputed Legacy, 2846-1990 (Hertingfordbury, U.K.: Roxford Books,
1991), pp. 131-136; and Ganguly, Origins of War in South Asia, pp. 13-14.
72. The arrival in 1983 of Muslim refugees fleeing a massacre in Assam a n d t h e assault in 1984
by Indian forces on the Sikh Golden Temple in Amritsar did not help either. Sumit Ganguly, The
Crisis in Kashmir: Portents of War, Hopes for Peace (Washington, D.C.: Woodrow Wilson Center Press,
1993, pp. 7&90; and Lamb, Disputed Legacy, pp. 327-330.
73. Ganguly, Crisis in Kashmir, pp. 107-108,133, 152-156; "India and Pakistan Plan Kashmir Talks,"
New York Times, June 24,1997. In 1998, in the first free elections in Kashmir in nearly two decades,

When All Else Fails 1 143

The most important conflict within Pakistan since independence has been
the secession of Bengali-speaking East Pakistan (Bangladesh) from the mainly
Urdu-speaking West in 1971 (which also caused the 1971 international war
between Pakistan and India).74Possible explanations of this conflict include
undemocratic institutions that allowed West Palustan to dominate Pakistani
politics, reaction against state repression, and ethnic and linguistic tensions;
however, the conflict cannot be charged to the 1947 separation of Hindus and
Muslims.
Finally, both Pakistan and India contain numerous additional ethnic minority groups, including Sikhs, Nagas, Tripuras, and others within India, and
Baluchis, Pathans, and Mohajirs in Pakistan, that have agitated for greater
autonomy or even rebelled.75Some of these disputes predate independence,
and we have no way to determine whether any of them would have occurred
in the context of a united India. There is one exception; the conflict that has
emerged in recent years, mainly in the city of Karachi, between Mohajirs
(refugees from India and their descendants) and the pre-1947 Sindhi community is chargeable to the partition of India and resulting population exchanges,
and has cost more than 3,500 lives as of 1997.76
WHAT SHOULD HAVE BEEN DONE? The problem with Indian independence
was not partition, but that partition did not go far enough. First and most
important, there was no provision for a Sikh homeland, either as an independent state or as a province of India. Even though Sikhs were only 1.2
percent of the population of India, 6 million people as determined, organized,
and armed as they were could not be ignored and still hope for peace. The
hard part is that because Sikhs were not an absolute majority in any district of
Punjab, any homeland would have required planning for substantial population transfers and therefore substantial commitment of resources for refugee
protection and resettlement.
Second, Kashmir should have been included in the general settlement, regardless of the maharaja's wishes. The result would likely have been a partition
more favorable to Pakistan than the one achieved by war, and would have

the antiseparatist National Conference won 4 of 6 seats, while turnout was low as a result of a
boycott called by Muslim separatists. Surinder Oberoi, "Three Die and 12 Abducted during
Kashmir Elections," AAP Information Services, March 1, 1998; and "Betrayal in Jammu and
Kashmir," The Hindu, April 5,1998.
74. Ganguly, Origins of War in South Asia, p. 58.
75. Myron Weiner, Sons of the Soil: Migration and Ethnic Conflict in lndia (Princeton, N.J.: Princeton
University Press, 1978).
76. Reuters World Service, January 28, 1995; and Reuters North American Wire, April 28,1997.

lnternational Security 23:2 244

avoided stranding a large community, both vulnerable and threatening, on the


wrong side of the line.
PALESTINE A N D ISRAEL

The War of Israeli Independence cost the lives of 6,000 Jews and probably more
than 10,000 Arabs, and displaced well over a million people; about 750,000
Palestinian Arabs fled Israel, while over half a million Jews migrated from Arab
countries to Israel.77As in Ireland and India, these costs were results of security
dilemmas generated by independence, not of partition.
DID PARTITION REDUCE VIOLENCE OR INCREASE IT? Neither; it had no effect.
Very few people today would suggest that the partition of Palestine could have
been avoided. From 1946 onward, the Jewish population mounted a revolt that
British forces could not control, forcing their withdrawal. A unified independent Palestine was impossible because the Jews would not submit to rule
by the Arab majority, and the Arabs would not accept any arrangement that
allowed for political power for Jews or even continued Jewish immigration.
Palestinian Arabs staged major riots in 1929 over these issues, as well as a
major rebellion from 1936 to 1939.
The departure of British power from Palestine, competing Arab and Jewish
land claims, and the intermixed population settlement pattern created a security dilemma so intense that civil war was certain. The three main areas of
Jewish settlement-astern
Galilee, the coastal strip from Haifa to Tel Aviv, and
Jerusalem-all contained substantial Arab populations and, even more important, were separated from each other by all-Arab regions (see Map 3). A
partition plan that did not envisage substantial population transfers could have
done nothing to resolve this security dilemma, and the one voted by the United
Nations in November 1947 was no exception.78
Faced with armed conflict internally as well as external invasion, the Jewish
state could survive only if it could (1) expand to link together its three main
parts, and (2) move out a large fraction of the Arabs from Jewish-controlled
77. Chaim Herzog, The Arab-Israeli Wars (New York: Random House, 1982), p. 108; Walid Khalidi,
All That Remains: The Palestinian Villages Occupied and Depopulated by Israel in 1948 (Washington,
D.C.: Institute for Palestine Studies, 19821, p. 582; and Joseph P. Schechtman, The Refugee in the
World: Displacement and Integration (New York Barnes; Yoseloff, 1964), p. 262.
78. The plan divided Palestine into eight parts: three main Jewish and three main Arab enclaves,
meeting at two points in such a way that none of the Jewish or Arab enclaves were contiguous.
A seventh small Arab enclave (the city of Jaffa) was surrounded by Jewish territory and the eighth
("internationalized" Jerusalem) by Arab territory. T.G. Fraser, Partifion in Ireland, India, and Palestine
(New York: St. Martin's Press, 1984), pp. 177-183.

W h e n A l l Else Fails 145

Map 3. Palestine, 1946.

LEBANON

SOURCE:

Supplement to a Survey of Palestine (Jerusalem: Government Printer, 1947).

International Security 23:2 146

areas, especially from the most strategically critical locations. The pattern of
ethnic cleansing during the war followed security dilemma logic. Israeli government leaders and military forces in some places encouraged Arab inhabitants to remain, in others harassed or frightened many into flight, and in yet
others carried out forced expulsions, depending on the strategic needs at each
place and time.
Two polar opposite events that took place just ten weeks and fifty miles apart
illustrate this dynamic. In late April 1948 the Jews gained control of the city of
Haifa, whose population was roughly evenly mixed. Almost immediately the
entire Arab population, not just of Haifa but of all the coastal towns north to
the Lebanese border, abandoned their homes and fled to Lebanon. The Israeli
government was surprised and dismayed by this exodus, and Israeli politicians
and soldiers tried to persuade the population to stay.79In contrast, in early July
Haganah troops surrounded the Arab towns of Lod (Lydda) and Ramle southeast of Tel Aviv and expelled their entire populations40,OOO in a l l - o n fortyeight hours' notice, with only what property they could carry?'
The difference in strategic need accounts for this difference in behavior. The
Mediterranean coast north of Haifa was an all-Arab region, with no Jews living
in or beyond it who might need to be rescued. Lod and Ramle, however, sat
astride the main Tel Aviv-Jerusalem road at a time when the Jewish portion
of Jerusalem was under siege and Israeli leaders were uncertain whether it
could hold out. Resupply and reinforcement convoys had to travel through the
streets of the two towns, and routinely had to fight their way through. Jerusalem could not be secured as long as these towns (and certain other villages
further along the route) remained in Arab hands."
The war ended in 1949 when the Israeli-Arab security dilemma was ameliorated, if not fully resolved, by nearly complete separation of the two communities. Israel secured a defensible territory that included the major Jewish
settlements and the spaces between them. The remaining 156,000 Arabs made
up no more than 15 percent of the new state's population and were disorganized and demoralized, and therefore were not perceived by Jewish Israelis as
a significant threat.82Gaza, the West Bank, and the Old City of Jerusalem came
79. British District Police Reports, April 26 and 28, 1948, cited in Schechtman, The Refugee in the
World, p. 191.
80. Benny Morris, "Operation Dani and the Palestinian Exodus from Lydda and Ramle in 1948,"
Middle East journal, Vol. 40, No. 1 (Winter 1986), pp. 82-109.
81. Jewish terrorists not under government discipline carried out an even more brutal cleansing
operation in the village of Deir Yassin, which also overlooks the Jersualem road.
82. Howard M. Sachar, A History of Israel (New York: Knopf, 1979), pp. 382, 395.

When All Else Fails I 147

under control of the Egyptian and Jordanian armies; no Jews remained in these
areas.
DID PARTITION CAUSE INCREASED HATRED AND GENERATE NEW CONFLICTS?

Since independence Israel has not experienced significant internal intergroup


violence, but has fought four wars against neighboring states in 1956, 1967,
1969-70, and 1973; has occupied parts of Lebanon twice, in 1978 and 1982-85;
and since 1987 has faced organized resistance to its occupation of the West
Bank and Gaza. Israel has been the target of continuing terrorism, and Israeli
citizens have carried out terrorist attacks as well.
Partition played no role in causing the continuation of violence after 1949,
which resulted simply from the presence of the Jewish state. Most Arabs
remained unwilling to accept the permanence of a large Jewish presence in
Palestine, while Israelis suspected the Arabs of genocidal aims. However,
despite wars and terrorism, most civilians on both sides have been safe most
of the time for nearly fifty years, which was not true before the population
transfers. Over time, many on both sides have abandoned the most extreme
enemy images. Israel has signed peace treaties with two of its neighbors, and
enjoys calmer relations with most other Arab states than it once did.
The one new conflict generated since Israel's independence, the Palestinian
infifada, was caused in large part by Israel's policy of planting Jewish settlements in the West Bank and Gaza, in effect remixing populations that had been
separated. Prior to the coming to power of the Likud Party in 1977, most
settlements were placed either just across the 1949-67Green Line or as security
outposts in remote areas. After 1977, however, the pace of construction accelerated, including establishment of settlements deep in the midst of regions
populated by Arabs, most infamously the placement of 400 Jews in the middle
of the 100,000-population city of Hebron. This generated a new security dilemma, both because the new settlements consumed more and more land and
because provisions for their security required restricting freedom of movement
between Arab towns."
WHAT SHOULD BE DONE? Mutual security for Jews and Palestinians requires
once again substantially separating the populations by removing those Jewish
settlements furthest from Israel proper that cannot be maintained except by
continuing military suppression of the Palestinians. This includes the three

83. Ze'ev Schiff and Ehud Ya'ari, Intifada: The Palestinian Uprising-Israel's Third Front (New York:
Simon and Schuster, 1989); and Geoffrey Aronson, Creating Facts: Israel, Palestinians, and the West
Bank (Washington, D.C.: Institute for Palestine Studies, 1987).

International Security 23:2 248

settlements in Gaza, Hebron, and dozens of others. It is not necessary that all
Jewish settlements be removed; many that are sited directly across the Green
Line from Israel proper and serve as suburbs to Jerusalem or to the coastal
plain could be incorporated into Israel without also including Arab population
centers. Where populations are already separated, it is easier to move borders
than people.84Some might object to such a unilateral border adjustment on
grounds of equity or law. It would, however, make both sides safer and might
also help them muster the political will to substantially implement the 1993
Oslo peace agreement.
CYPRUS

Critics of the 1974 de facto partition of Cyprus argue that even though there
have been virtually no casualties since then, the partition and population
exchange have actually made the conflict worse, not better: The division of
Cyprus is little more than a long standoff that remains volatile and continues
to require the presence of U.N. troop^."'^
In fact, however, the situation has remained remarkably stable since 1974.
There have been only twelve deaths in ethnic strife on the island in twenty-four
years.% This generation of calm compares starkly to the escalating ethnic
violence on Cyprus from 1955 to 1974.
DID PARTITION REDUCE VIOLENCE OR INCREASE IT? Prior to 1960 Cyprus was
under British rule; the population was approximately 80 percent Greek and 20
percent Turkish. In the 1950s the main Cypriot independence movement,
EOKA, was a specifically Greek movement whose aim was union with metropolitan Greece (enosis). The Turkish community, fearing Greek domination,
preferred continued colonial rule or, if independence could not be avoided,
~artition.~~

84. Alexander B. Downes, The Holy Land Divided? Theory and Practice for a Successful Partition

of Palestine, unpublished ms., University of Chicago, suggests that Jews now living in the West
Bank exchange places with the Palestinians of Gaza. This seems to me excessive.
85. Kumar, Troubled History of Partition, p. 29. Kumar further claims that there was a war scare
between Greece and Turkey over Cyprus as recently as August 1996, but this actually amounted
to no more than some moderately warm rhetoric by Greek Prime Minister Costas Simitis, plus
Turkish protests about attacks on three consulates. See Michele Kambas, Greek PM Slams Turkey
on Arrival in Cyprus, Reuters European Community Report, August 17,1996.
86. Five in 1975, two between 1989 and 1993, and five in 1996. List of Deaths on Green Line since
1974, Agence France-Presse, August 11,1996; Patrick Baz, Agence France-Presse, August 14,1996;
and Michele Kambas, Reuters World Service, October 15, 1996.
87. Nancy Cranshaw, The Cyprus Revolt: A n Account of the Struggle for Union with Greece (London:
George Allen and Unwin, 19781, pp. 42-50,62-67,71-75; Christopher Hitchens, Hostage to History:

When All Else Fails 1 149

The population settlement pattern on the island contributed to an intense


security dilemma. Although most villages and quarters of major towns were
populated exclusively by one group, Greek and Turkish settlements were
spread throughout the island, with only a slight bias of Turkish concentration
toward the north.@ As a result, from 1955 to 1974 Cyprus underwent four
major rounds of civil war. First, starting in 1955 EOKA attacked British forces,
Greek collaborators and communists, Turkish Cypriots serving in the British
government and police, and increasingly the Turkish community at large;
Turkish terrorist groups attacked Greek civilians. At least 509 people died
before Britain granted independence in 1960.89
Second, although the new states constitution incorporated power-sharing
principles and prohibited enosis, Greek Cypriot leaders, including President
Archbishop Makarios, continued to advocate Greek majority rule as well as
enusis. Governance was soon paralyzed by obstructive tactics on both sides,
and in 1963 Greek Cypriots abrogated the constitution and established a new,
all-Greek government. In December 1963 civil war broke out again, and Greek
forces soon controlled nearly the whole country except for a few towns in the
northern part of the island (see Map 4). In August 1964 an offensive against
the Turkish Cypriots only remaining outlet to the sea was stopped by the
Turkish Air Force. At least 550 people died and 25,000 Turkish Cypriot refugees
were resettled. From 1964 to 1968 most of the Turkish-held enclaves were
under a de facto economic blockade.
Third, after a military buildup by Greek Cypriot nationalists, including
12,000 Greek Army troops deployed on Cyprus as a deterrent to further
Turkish interference, Greek Cypriot forces again attacked Turkish villages in
April and November 1967. Turkey responded by threatening air strikes and
troop landings. This time a deal brokered by the United States and the UN led
Cyprusfrom the Ottoman Empire to Kissinger (New York Farrar, Straus, and Giroux, 19891, pp. 4246;
and Tozun Bahcheli, Greek-Turkish Relations since 2955 (Boulder, Colo.: Westview Press, 1990),
pp. 2&30.
88. Richard A. Patrick, Political Geography and the Cyprus Conflict: 1963-1971 (Waterloo, Ontario:
University of Waterloo, 19761, pp. 8-11,
89. British official figures, reported in Cranshaw, Cyprus Revolt, p. 406.
90. Makarios was, however, prepared to be patient on the latter, because he did not want to
provoke Turlush intervention. Cranshaw, Cyprus Revolt, pp. 341-345, 366-367; P.N. Vanezis,
Makurios: Pragmatism v. Idealism (London: Abelard Schuman, 1974), pp. 12S133; and Stanley
Mayes, Makarios: A Biography (New York: St. Martins Press, 1981), pp. 159-166.
91. Patrick, Political Geography and the Cyprus Conflict, p. 46; and Cranshaw, Cyprus Revolt, pp. 367373. The presence of 7,000 UN peacekeepers during the latter part of the fighting had little effect.
92. Makarioss personal interest in enosis waned after the April 1967 military coup in Athens, but
he felt unable to oppose the projects of Greek Cypriot nationalists, which commanded great

e Turkish Cypriot Enclaves in the Late 1960s.

M e d l r e r r e n e

S e .

re Oberling, The Road to Bellepais: The Turkish Cypriot Exodus to Northern Cyprus (New York: Columbia
ress, 1982). Used by permission.

When All Else Fails 252

to the withdrawal of the Greek Army units, and in 1968 the movement restrictions on Turkish Cypriots were lifted. Deaths between 1964 and 1968 were
probably about 600?3 From 1968 to 1974 the sides conducted direct negotiations, but the Greek Cypriots would not give up the goal of enosis, and the
Turks would not accept it.94
Finally, in 1971 a new Cypriot nationalist organization, EOKA B, which was
supported by the military junta ruling in Athens, began terrorist attacks. On
July 15, 1974, Makarios was overthrown in a bloody coup.95An unknown
number of people, probably in the hundreds, were killed. Nicos Sampson, an
ultra-rightist infamous for heading massacres in 1963-64, was appointed president. A massive islandwide pogrom appeared imminent.96
Turlush forces then invaded Cyprus and overran 37 percent of the island,
creating a Turkish-controlled zone in the north. Sampsons government collapsed and Makarios resumed power. Approximately 1,000 people, mostly
Cypriot National Guardsmen and Greek civilians, were killed in this operation,
not many less than the death toll of the previous twenty years. The invasion,
however, did save thousands who would likely have been murdered if men
like Sampson had actually executed their program. Approximately 200,000
Greek refugees moved south of the line, and about 60,000 Turkish Cypriots
moved north.97Even though the Turkish invasion was motivated in part by
nationalist rather than humanitarian concerns, it saved lives; fortunately, we
will never know just how many.

popularitv. In 1970 he barely survived a nationalist assassination attempt. Mayes, Makarios,


183-f86, 206-207.
93. Hitchens, Hostage to History, p. 65; Bahcheli, Greek-TurkishRelations, pp. 73-74,173; and Patrick,
Political Geopraahw and the Cwrus Conflict, D. 119.
94. For details of the negotiations akd iiternal politics of the sides, see Polyvios G. Polyviou,
Cyprus: Conflict and Negotiation, 1960-1980 (New York Holmes and Meier, 1980), pp. 62-132; and
Bahcheli, Greek-Turkish Relations, pp. 89, 167, 175-188.
95. Although EOKA B was an illegal organization, it commanded great popular support among
Greek Cypriots, including much of the Cypriot National Guard. The funeral in January 1974 of its
founder, General George Grivas, was attended by one-fifth of the entire Greek population. Polyviou, Cyprus, pp. 120-130; and Pierre Oberling, The Road to Bellepais: The Turkish Cypriot Exodus to
Northern Cyprus (New York Columbia University Press, 19821, pp. 149-150.
96. Sampson was elected to the Greek Cypriot House of Representatives in 1969 on the slogan
Death to the Turks! Oberling, Road to Bellepais, p. 160. According to Makarios, Sampson and the
head of the Athens junta in 1974, General Dimitrios Ioannides, had come to see him back in 1964
and proposed: Your Beatitude, here is my project. To attack the Turkish Cypriots suddenly,
everywhere on the island, and eliminate them to the last one. Quoted in Orianna Fallaci, Intervim
with History (New York Liveright, 1976), p. 318.
97. Cranshaw, Cyprus Revolt, p. 395; and Polyviou, Cyprus, p. 203.

p<

_I,

lnternational Security 23:2 1 152

DID PARTITION INCREASE HATRED AND GENERATE NEW CONFLICTS? Greek


Cypriots still have not accepted the partition of Cyprus, and no nation except
Turkey recognizes the Turkish Republic of Northern Cyprus. Several rounds
of negotiations have been held in which the Turks have offered to return some
land or to reunite Cyprus as a loose federation, but these proposals have not
been acceptable to the Greek side.
Despite the lack of reconciliation, the situation has remained calm as well as
safe for a generation. Both sides know that the Turkish Army garrison in the
north could defeat any possible irredentist offensive, and neither side has
shown any evidence of plans to try to disrupt the existing situation.

The Politics of Successor States


Critics of ethnic population transfers and partitions have overestimated the
risks that these remedies pose to political development. There are two main
areas of development that should be considered: democratization and treatment of ethnic minorities.
The four partitions studied in this article have produced nine de jure or de
facto successor states: the Republic of Ireland, Northern Ireland, India, Pakistan, Bangladesh, Israel, the Palestinian Authority, Cyprus, and the Turkish
Republic of Northern Cyprus (TRNC).The degree to which the political institutions of these states qualify as democratic is most easily measured by the
extent to which they hold periodic free elections that can alter the composition
of the government and its policies, and whose results are not overturned by
force, such as military coups.98 Of the nine successor states, five (Ireland,
Northern Ireland, India, Israel, and Cyprus) have political institutions that
clearly meet democratic qualifications. Four (Pakistan, Bangladesh, the Palestinian Authority, and the TRNC) do not fully qualify, but even these states are
not clearly less democratic than their pre-partition predecessors, and all are at
least as democratic as most of their neighbors.
Although Pakistan and Bangladesh have always had nominally democratic
forms, each has been under military rule for slightly less than half the time
since their respective dates of independence (Pakistan 1958-71 and 1977-88,
and Bangladesh 1975-86). In comparison, British colonial rule in India had
become, by the time of partition, largely democratic at the province level,
98. Samuel P. Huntington, The Third Wave:Democratization in the Late Twentieth Century (Norman:
University of Oklahoma Press, 1991), pp. 5-13.

When All Else Fails 1 253

although not at the national level and not in the princely states. Both Pakistan
and Bangladesh have democratic qualifications that are at least as strong as
those in any state in South, Southeast, or Central Asia, or in the Middle East.
Three exceptions are India and Israel (also partition successor states), and
arguably Sri Lanka.
Although it did not emerge until after the 1993 Oslo agreement, the Palestinian Authority can be considered as a de facto successor state of the partition
of the Palestine mandate. While the Authority has nominally democratic forms,
it has not yet had a change of leadership, and political expression is often
suppressed. It does not, however, compare unfavorably to the British mandatory government, which had no representative institutions above the level of
municipalities, or with other Arab ~tates.9~
We could also ask whether the 1947
partition affected the political development of adjacent states that gained
territory or absorbed refugees. There is little evidence that it did, as none of
these states except Lebanon had democratic institutions either before or after
partition."'
The Turkish Republic of Northern Cyprus also has democratic forms but has
never had a change of government. However, prior to partition in 1974 united,
formally democratic Cyprus never had a change of government either, and
allowed the Turkish minority no political role."' Compared with the democratic qualifications of its neighbors around the eastern Mediterranean, those
of the TRNC rank about average.
Even where there are democratic forms, we must also ask whether minorities
are nevertheless effectively disenfranchised. Four of these nine successor states
have significant minority populations. In the Republic of Ireland and in India,
minorities face no barriers to political participation, although the government
of India has proscribed secessionist parties that have threatened the territorial
integrity of the state.
Arab citizens of Israel have faced significant repression, most severely in the
first fifteen years after partition. Although many formal discriminatory policies
99. A British plan in the 1920s to create a legislature was blocked by Arab opposition. Bernard
Wasserstein, The British in Palestine: The Mandatory Government and the Arab-Jewish Conflict, 19171929 (Cambridge, Mass.: Basil Blackwell, 1978).
100. The presence of refugees from Palestine had some impact on the civil war from 1975 onward,
although the war was more the result of tensions among the native Christian, Sunni, and Shiite
populations. David Gilmour, Lebanon, the Fractured Country (New York St. Martin's Press, 1984).
101. Since 1974 the (Greek) rump state of Cyprus has had several changes of government in free
elections. Eric Solsten, Cyprus: A Country Study (Washington, D.C.: Government Printing Office,
1993).

International Security 23:2

I 154

have been lifted, until recently Arabs were barred from effective political
influence by an informal understanding that parties dependent on Arab votes
could not be part of a governing coalition."' In the last several years, however,
this barrier has weakened. The Labor Party-led coalition that governed Israel
from 1992 to 1996 depended on the support of two Arab parties, and Arab
votes nearly decided the 1996 prime ministerial contest between Shimon Peres
and Benjamin Netanyahu, in which Netanyahu led by 11percent among Jewish
voters but won the election by a margin of just 0.9 percent.'03
In Northern Ireland, Catholics faced few barriers to participation in nationallevel politics, but for five decades their influence on the province government
and local bodies was reduced by gerrymandering, especially in the cities of
Londonderry and Belfast, as well as in certain rural areas where Catholics were
in the majority. In addition, a 1946 "Representation of the People Bill" disenfranchised certain categories of potential voters, disproportionately Catholics.
These abuses were largely ended by the imposition of direct British rule in
1972.'04
Although these restrictions on minority rights are serious, repression of
ethnic minorities would likely have been worse in each case had partition and
population transfers not occurred. Even if we imagine that it could have been
possible in 1947 to construct a government of a united Palestine that would
have treated the Jewish minority more mildly than Arabs have fared in Israel,
the Jewish community saw the prospect of Arab rule as so dangerous that it
could have been imposed only by crushing the Jews' capacity to resist. Whatever tendencies toward tolerance such a government might have had initially
would surely have been undermined by the violence and mutual security fears
generated in the process of establishing control. The problem of the Protestants
in a united Ireland would have been the same.105

102. Ian Lustick, Arabs in the Jewish State: Israel's Control of u Nutional Minority (Austin: University
of Texas Press, 1980).
103. "Rabin Meets Hadash, DAP Representatives: Shas Hints at Abstention Tomorrow," Jerusalem
Post, November 9, 1993; and "Netanyahu Wins Israeli Election," Associated Press, May 31, 1996.
104. Schaeffer, Warpaths, pp. 166-167.
105. Critics also charge that partition successor states discriminate against minorities through
citizenship and language laws-for example, Israel's Law of Return, which grants automatic
citizenship to Jewish immigrants, and its 1950 Absentee Property Law, which bars the return of
Arabs who left the country during the 1947-49 war. However, official languages, citizenship laws,
and immigration practices that favor the majority are actually features of most nation-states,
including such liberal democracies as Norway, Denmark, Sweden, Germany, France, and Italy.

When All Else Fails I 155

Challenges for Separation and Partition


This analysis suggests three lessons for management of ethnic civil wars. First,
we need to identify the threshold of intergroup violence and mutual security
threats beyond which we must resort to separation and partition, and we
should set the threshold conservatively-no one wants to dissolve diverse
societies, even deeply troubled ones, that have any hope of avoiding massive
violence and attaining civil peace. The theory and evidence presented here can
help us identify cases that are clearly over that threshold; but absent a fully
developed theory of the causes of interethnic hostility in peacetime, we cannot
know exactly where the true threshold may lie.
Even this limited knowledge, however, is of policy use. We should not fail
to separate populations in cases that have already produced large-scale violence and intense security dilemmas, even if in some such cases we might later
wish that we had acted sooner and in yet other cases we may not be able to
decide whether to act.
Second, while the findings here do not suggest whether partitions of sovereignty should occur more or less frequently, they do imply that partition
should never be done unless the national communities are already largely
separate or will be separated at the same time. Partitions that do not unmix
hostile populations actually increase violence, as they did in Northern Ireland,
Kashmir, and Palestine, and when Croatia and Bosnia seceded from Yugoslavia. Defensible boundaries are also essential; the UN partition plan for Palestine, which gave each side three disconnected patches of land, could only
generate a bloody civil war.'"
In all four cases discussed above, ethnic separation reduced violence. Where
populations were largely separated to begin with, violence was much less
intense than where they were tightly intermingled. When warring populations
were separated, either by planned transfers or by ethnic cleansing, violence
subsequently declined. Except in those regions where hostile communities
remain mixed on the ground, all four cases have had less violence since
partition than before, and all sides have lower expectations of future violence
now than they did then.

106. Existing administrative unit boundaries are often given excessive weight, as in Kashmir in
1947 and Ulster in 1921. See Steven R. Ratner, "Drawing a Better Line: Uti Possidetis and the Borders
of New States," American Journal of International Law, Vol. 90, No. 4 (October 1996), pp. 59M24.

International Security 23:2 256

Finally, the record of twentieth-century population transfers and partitions


suggests major changes in how we treat refugees of ethnic wars. The international community should stop trying to prevent the movement of refugees
away from threats of ethnic massacres and should instead support and safeguard their resettlement. The UNHCRs policy of bringing safety to people,
not people to safety cannot be implemented in the midst of ethnic wars, and
attempts to do so are likely to cost the lives of some of the very people they
are supposed to save. Concern that facilitation of refugee movements amounts
to support for ethnic cleansing is misguided. Ethnic cleansing can only be
stopped by an army on the ground strong enough to defeat the cleansers.
Otherwise, making it harder for ethnic cleansers to expel their enemies only
invites them to escalate to murder.
Similarly, the international community should stop pressing winners of
ethnic wars to take back refugees of the other side, and should stop pushing
refugees to return when they fear for their lives if they do. After an ethnic war,
repatriating any substantial number of refugees back to territory held by the
other group risks making control of that territory once again uncertain, thus
re-creating the same security dilemma that helped escalate the conflict in the
first p1ace.O7
107. The severity of this risk depends on the robustness of the winning sides territorial control.
Nigeria could afford to reabsorb refugees of the Biafran war because the wars decisive outcome
meant that there was little chance of a new revolt. The Republika Srpska, which is far more fragile,
cannot. International demands that the Rwandan Patriotic Front accept the return of more than 2
million Hutu refugees placed the RPF in an impossible position. Because the RPF represented less
than 10 percent of the population, it could not claim legitimacy if it refused repatriation, but the
return of large numbers of Hutus, including adherents of the former government, was bound to
lead to a new round of civil war, and did. Hutu Rebels Terrorize Three Nations: The Slaughter
Continues, International Herald Tribune, January 29, 1998.

You might also like